Update app.py

app.py

@@ -1,34 +1,32 @@
-import os
 import gradio as ui
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 from threading import Thread
 
-# Mengambil token rahasia dari settings Space kamu
-hf_token = os.environ.get("HF_TOKEN")
+# Menggunakan Phi-3.5-mini-instruct (3.8B Parameter)
+model_id = "microsoft/Phi-3.5-mini-instruct"
 
-model_id = "Qwen/Qwen3-0.6B-Instruct"
-
-print("Memuat Tokenizer & Model Qwen3-0.6B menggunakan Akses Token...")
-# Kita masukkan parameter token=hf_token agar lolos verifikasi download
-tokenizer = AutoTokenizer.from_pretrained(model_id, token=hf_token)
+print("Memuat Tokenizer & Model Phi-3.5 ke RAM (Bakal butuh waktu agak lama)...")
+tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(
     model_id,
-    torch_dtype=torch.float32,
-    device_map="cpu",
-    token=hf_token
+    torch_dtype=torch.float32, # Stabil untuk kalkulasi CPU
+    device_map="cpu"
 )
-print("Qwen3 Berhasil Dimuat dan Siap Beraksi!")
+print("Phi-3.5 Berhasil Dimuat dan Siap Beraksi!")
 
-def chat_qwen(message, history):
+def chat_phi(message, history):
     conversation = []
     
+    # Menyusun riwayat percakapan terdahulu
     for user_msg, ai_msg in history:
         conversation.append({"role": "user", "content": user_msg})
         conversation.append({"role": "assistant", "content": ai_msg})
         
+    # Memasukkan prompt baru dari user
     conversation.append({"role": "user", "content": message})
     
+    # Mengonversi pesan sesuai dengan template chat khusus dari Microsoft Phi
     input_ids = tokenizer.apply_chat_template(
         conversation, 
         tokenize=True, 
@@ -36,16 +34,18 @@ def chat_qwen(message, history):
         return_tensors="pt"
     ).to("cpu")
     
+    # Menggunakan Streamer agar output teks muncul mengetik secara real-time
     streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
     
     generation_kwargs = dict(
         input_ids=input_ids,
         streamer=streamer,
         max_new_tokens=512,
-        temperature=0.6,
-        top_p=0.95
+        temperature=0.7,
+        top_p=0.9
     )
     
+    # Menjalankan inferensi di background thread agar UI tidak mengalami freeze/stuck
     thread = Thread(target=model.generate, kwargs=generation_kwargs)
     thread.start()
     
@@ -54,10 +54,11 @@ def chat_qwen(message, history):
         partial_text += new_text
         yield partial_text
 
+# Mengaktifkan antarmuka Chatbot menggunakan UI modern Gradio
 demo = ui.ChatInterface(
-    fn=chat_qwen,
-    title="🤖 Qwen3-0.6B Chatbot Authenticated",
-    description="Berhasil berjalan menggunakan akses token privat akunmu!",
+    fn=chat_phi,
+    title="🧠 Phi-3.5 Mini Instruct Chatbot",
+    description="Merasakan kekuatan model penalaran 3.8B buatan Microsoft, berjalan stabil langsung dari CPU Hugging Face Space Gratisan.",
     theme="soft"
 )